Summary of prospective cohort studies examining associations between free sugars exposures and adiposity in children
| First author (year), country, study name | Population and recruitment | Follow-up (years) | Exposure and assessment method | Outcome and assessment method | Covariates and stratification | Findings* | +/0/−† |
|---|---|---|---|---|---|---|---|
| Berkey (2004),75 USA, Growing Up Today Study | 1175 adolescents aged 9-14 years, whose mothers were in the Nurses’ Health Study II; consecutive sampling used in original Nurses’ Health Study II cohort | 1 | SSB intake (serving/day) and increase in SSB intake (1 serving/day increase); FFQ | Change in BMI (continuous); self reported | Adjusted for age, dietary intake, ethnicity, growth, physical activity, previous BMI z score, screen time, Tanner stage, total energy intake; stratified by sex | In boys, significant associations between SSB consumption at baseline and 1 year weight gain, and between increasing SSB consumption and 1 years weight gain were seen in adjusted analyses (both P≤0.038). After adjusting for total energy intake, effect sizes in boys and girls were reduced, and associations were rendered non-significant | + |
| Blum (2005),83 USA | 164 children aged 9.3 years (SD 1) attending elementary school; convenience sampling | 2 | SSB intake (ounces/day); 24 h recall | Change in BMI z score (continuous); measured | Adjusted for baseline BMI z score, intake of diet soda at follow-up | Regression analysis of BMI z score at follow-up showed no association with SSB consumption (data not reported) | 0 |
| Butte (2009),84 USA, Viva la Familia Study | 789 Hispanic adolescents aged 9-14 years, at least 1 overweight child in family; convenience sampling of families | 1 | Energy intake (%) from sucrose, and from added sugar; 24 h recall | Weight gain (kg) (continuous); measured | Adjusted for age, BMI status sex, Tanner stage | Energy intake from sucrose and added sugars was not found to be associated with weight gain | 0 |
| Dubois (2007),94 Canada, Longitudinal Study of Child Development in Québec | 380 children aged 2.5 years; random sampling | 2 | SSB intake between meals at age 2.5, 3.5, and 4.5 years, categorised as non-consumers, regular consumers, or other (including children who changed consumption); FFQ | Overweight BMI (≥95th percentile) at age 4.5 years (OR); measured | Adjusted for birth weight, dietary intake, maternal smoking, number of obese parents, physical activity, SES, sex, total energy intake | Regular consumers were 2.36 times more likely to be obese at age 4.5 years than non-consumers (95% CI 1.03 to 5.39; P≤0.05) | + |
| Faith (2006),81 USA | 825 children aged 1-5 years enrolled in the Special Nutrition Program for Women, Infants and Children in New York State; consecutive sampling | 2 | FJ intake (servings/day) ; questionnaire | Change in BMI z score (continuous); measured | Adjusted for baseline weight-for-height z score and weight status, dietary intake, parental feeding behaviour; stratified by being at risk of overweight (weight-for-height z score ≥85th percentile) and not being at risk of overweight | At follow-up, each serving/day of FJ intake at baseline was associated with a 0.005 increase in BMI z score (P<0.01) in pooled analysis, a 0.009 increase for people at risk of overweight (P<0.01), and a non-significant increase of 0.003 for those not at risk of overweight | + |
| Fiorito (2009),86 USA | 166 girls aged 5 years at baseline; convenience sampling | 10 | SSB intake (<1, ≥1 to <2, or ≥2 servings/day, and as a continuous variable); 24 h recall | BF (%), WC (cm), BMI percentile, overweight status (% overweight in each SSB intake group) (all continuous); measured | Adjusted for maternal BMI, parental education, SSB intake at 15 years, total energy intake (all adjusted for separately) | SSB intake at age 5 years was a significant predictor of adiposity at each 2 year time point during ages 5-15 years in unadjusted ANOVA analyses. Greater consumption of SSB at 5 years was associated with a higher WC at 15 years, after adjusting for intake of SSB and energy at 15 years and family income at 15 years (all P<0.05), but association was non-significant after adjusting for maternal BMI and parental education | + |
| Haerens (2010),77 Belgium, Longitudinal Eating and Activity Study | 585 children aged 10 years attending elementary school; random selection of schools, consecutive sampling of children | 4 | Intake (servings/week) of SSSD and sweets; FFQ | Change in BMI z score (continuous); self reported | Adjusted for dietary intake and change in dietary intake, SES, sex, time since baseline | No associations seen between consumption of SSSD/sweets and BMI z score | 0 |
| Herbst (2011),87 Germany, DONALD Study | 216 infants aged 1 year, born to German citizens; convenience sampling | 6 | Energy intake (%) from total added sugar, added sugar in beverages and sweets, added sugar from other sources, plus changes in intake between 1-2 years; 3 day diet record | BMI z score and BF (%) (both continuous); measured | Adjusted for birth weight, BMI at birth, breastfeeding, dietary intake, gestational age (weeks), paternal education, maternal overweight, sex | Higher % energy intakes from total added sugars and added sugar in beverages and sweets at age 1 year were negatively associated with BMI z score at age 7 years. BMI z score increase per 1% energy from total added sugar: β −0.116 (SE 0.057, P=0.04; from added sugar in beverages and sweets: −0.250 (0.103, P=0.02). No associations were observed between % energy intakes from sugars and % BF | — |
| Johnson (2007),98 UK, Avon Longitudinal Study of Parents and Children, Children in Focus | 521 children aged 5 years or 682 aged 7 years born in the final 6 month of the AVON study; consecutive population sampling used to obtain AVON cohort, random sampling within the cohort | 2/4 | Intake (servings/day) of SSB or FJ at 5 or 7 years; 3 day diet record | Fat mass index (kg/m5.8)‡ and change in fat mass (kg) (both continuous); measured | Adjusted for baseline BMI, dietary intake, height, parental BMI and SES, sex, TV viewing | After adjusting for sex and height, fat mass index‡ at age 9 years correlated with SSB intake at 5 and 7 years (Pearson’s correlation coefficient P values 0.06 and 0.03, respectively). In multivariate regression analysis, fat mass at 9 years was negatively associated with FJ intake at 5 years (−0.55 kg/daily serving, 95% CI –1.08 to –0.02) | + (SSB), − (FJ) |
| Libuda (2008),88 Germany, DONALD Study | 235 children aged 9–13 years, born to German citizens; convenience sampling | 5 | Energy (MJ) derived from SSB, FJ, energetic beverages (SSB and FJ); and change in intake; 3 day diet record | Change in BMI z score (continuous) and % change in BF (continuous); measured | Adjusted for age, birth weight, energy intake from other sources and change in energy intake from other sources, maternal education and BMI, time, years of adolescence; stratified by sex | 5 year changes in BMI z score and BF (%) were not related to baseline intake of SSB, FJ, or combination of both. Change in BMI z score was positively associated with concurrent change in energetic beverage intake in girls (β 0.07, P=0.01) and with concurrent change in FJ intake in girls (β 0.096; P=0.013) | 0/+ |
| Lim (2009),97 USA, Detroit Dental Health Project | 365 children aged 3-5 years from areas in Detroit with the highest proportion of low income households; random sampling | 2 | SSB (all, fruit drinks and SSSD) intake (ounces/day) and change in intake (decrease, increase, no change) FFQ | Change in BMI z score (continuous) and incident overweight (BMI ≥85th percentile) (OR); measured | Adjusted for baseline BMI, caregiver’s BMI, SES, total energy intake | Odds of becoming overweight increased by 4% (95% CI 1% to 7%; P<0.05) per ounce of fruit drink consumed per day at baseline, 4% (1% to 6%; P<0.05) per ounce of all SSB consumed per day at baseline, and 4% per ounce of SSSD consumed per day at baseline, although non-significantly (95% CI −1% to 10%; P>0.05). No significant association found between change in beverage intakes and incidence of overweight (data not reported). No significant association found between change in beverage intakes and change in BMI z score | + |
| Ludwig (2001),95 USA, Planet Health | 548 multiethnic children aged 11-12 years participating in an intervention study as controls; convenience sampling of schools, random assignment of schools to intervention/control, consecutive sampling of children | 1.6 | SSB intake (servings/day) and change in intake (increase of 1 serving/day); FFQ | Incident obesity (BMI and triceps skinfold ≥85th percentile) (OR) and change in BMI (continuous); measured | Adjusted for age, baseline BMI and triceps skinfold, dietary intake and change in dietary intake, ethnicity, physical activity, school indicator variables, sex, total energy intake, TV viewing, and change in TV viewing | For each serving of SSB consumed/day at baseline, BMI increased by 0.18 (95% CI 0.09 to 0.27; P=0.02), and for each SSB serve/d increase from baseline, BMI increased by 0.24 (0.10 to 0.39; P=0.03). Incident obesity was not associated with baseline SSB intake (OR 1.48, 95% CI 0.63 to 3.47; P=0.27), but was associated with an increase in SSB intake, with a 60% higher risk of developing obesity for each serving/day increase (95% CI 14–124%; P=0.02). | + |
| Nissinen (2009),82 Finland, Cardiovascular Risk in Young Finns Study | 2139 children aged 3, 6, 9, 12, 15, and 18 years; random sampling | 21 | Intakes of SSSD or sweets in childhood (per 10 units/month), increase, decrease, or no change in SSSD/sweets intake from childhood to adulthood; questionnaire | Adult BMI (continuous) and adult overweight status (BMI ≥25) (OR); measured | Adjusted for adult education, age, overweight status in childhood, physical activity, smoking; stratified by sex | The predicted difference in adult BMI per 10-unit consumption difference of SSSD or sweets in childhood did not reach statistical significance for males or females. An increase in the frequency of SSSD consumption from childhood to adulthood was associated with a higher BMI in adult females (β 0.45, SE 0.12; P<0.001). Adult male overweight status was not associated with change in intake of SSSD or sweets from childhood to adulthood. Adult female overweight status was not associated with change in intake of sweets, however, compared to females who maintained a low intake of SSSD from childhood to adulthood, those who increased intake were 1.9 times more likely to be overweight as adults (95% CI 1.38 to 2.61) | 0/+ |
| Phillips (2004),89 USA, Massachusetts Institute of Technology Growth and Development Study | 132 premenarcheal girls aged 8-12 years attending public schools in Massachusetts; convenience sampling | About 7 | Energy intake (%) from energy dense snacks (SSSD, candy/chocolate, baked goods, ice cream), divided into groups according to intake level; FFQ | BMI z score and BF (%) 4 years after onset of menarche (both continuous); measured | Adjusted for age at menarche, dietary intake, parental overweight | Of 5 categories of energy dense snacks, only SSSD intake at baseline was significantly related to BMI z score 4 years after menarche, although the effect size was small (compared with the group with the lowest intake, β 0.172 for the group with the second highest intake, and β 0.178 for the group with the highest intake; both P<0.001). No associations were seen between intake of energy dense snack and BF (%) | + |
| Skinner (2001),90 USA | 72 white children aged 2 years, born to parents with middle to high SES; convenience sampling | 4 | Longitudinal intake of FJ (ounces/day; mean across 7 sets of measurements); 24 h recall and 3 day diet record | BMI and ponderal index (kg/m3) (both continuous); measured | Adjusted for age, baseline BMI or ponderal index, longitudinal total energy intake, parental BMI, sex | Longitudinal FJ intake was not significantly associated with BMI (β −0.057; P=0.099), but was negatively associated with ponderal index (β −0.065; P=0.05) | 0/− |
| Stoof (2011),91 Holland, Amsterdam Growth and Health Longitudinal Study | 238 adolescents aged 13 years, attending secondary schools of middle to high SES in and around Amsterdam; convenience sampling | 24-30 | Intake of SSB (servings/day); diet history | BMI, BF (%) and trunk fat (%) (all continuous); measured | Adjusted for age, baseline BMI, developmental age at baseline, physical activity, total energy intake | For men, each additional daily serving of SSB at age 13 years was associated with greater BF (%) in adulthood (β 1.14, 95% CI (0.04 to 2.23); P=0.04). In women, each additional daily serving of SSB at 13 years was associated with greater trunk fat in adulthood (β 1.62 (0.14 to 3.10); P=0.03). No relation between SSB consumption at age 13 years and BMI in adulthood found in either sex | + |
| Striegel-Moore (2006),92 USA, National Heart, Lung and Blood Institute Growth and Health Study | 2371 non-Hispanic black and white girls aged 9-10 years, attending schools in Richmond, Hamilton County, and Maryland, from families enrolled in a health maintenance organisation in the Washington, DC area, and girl scouts in the Washington, DC area; convenience sampling | 10 | Intake of SSSD, FJ, fruit drinks (100 g/day); 3 day diet record | Concurrent change in BMI (continuous); measured | Adjusted for ethnicity, intake of other beverages, study site and visit, total energy intake, within-individual correlation of repeated measures | Beverage intake was measured at baseline and at most annual follow-up visits (visits 1-5, 7, 8, and 10). Concurrent change in BMI was positively associated with intake of SSSD per 100 g/day (β 0.011 (SE 0.005); P<0.050). BMI was not found to be associated with intake of FJ or fruit drinks (both P≥0.05) | + (SSB), − (FJ) |
| Vanselow (2009),78 USA, Project EAT | 2294 adolescents aged 11-15 years, attending 31 ethnically and socioeconomically diverse schools in Minneapolis or St Paul; convenience sampling | 5 | Servings/week of SSSD, fruit drinks, orange juice, apple juice; FFQ | Change in BMI (continuous); self reported | Adjusted for age, physical activity at baseline and follow-up, baseline BMI, cohort, ethnicity, other beverage intake, SES, sex, TV viewing at follow-up | No association found between consumption of SSSD, fruit drinks, orange or apple juice at baseline and BMI at follow-up | 0 |
| Viner (2006),79 UK, 1970 British Birth Cohort | 4461 adolescents aged 16 years, born in one week in 1970 in England, Northern Ireland, Scotland, and Wales; consecutive sampling | 14 | 0, 1, or ≥2 servings of SSSD on previous day; questionnaire | Change in BMI z score at age 16-30 years (continuous); self reported | Adjusted for baseline BMI z score, height at baseline and follow-up, SES, sex | Compared with people who consumed no SSSD on the day before baseline, those who reported consuming ≥2 servings had a significantly greater change in BMI z score over the 14 year follow-up (β 0.13 (95% CI 0.01 to 0.26); P<0.04) | + |
| Weijs (2011),80 Holland | 120 infants aged 4-13 months; convenience sampling | 8 | Intake of total sugar and intake of beverage sugar (% of total energy, g/day); 2 day diet record | Overweight status at age 8 years (≥1 increased in BMI z score) (OR); self reported | Adjusted for animal protein intake, baseline age, baseline body weight, breastfeeding, SES, sex | In unadjusted analyses, sugar intake at baseline (% of total energy and g/day) and beverage sugar intake at baseline (g/day) did not differ between people not overweight and those overweight at follow-up. However, beverage sugar intake (as % of total energy) differed between the groups (P=0.04). For intake of beverage sugar per 1% of total energy intake, adjusted OR for overweight at age 8 years was 1.13 (95% CI 1.03 to 1.24) | + |
| Welsh (2005),96 USA, Pediatric Nutrition Surveillance System and Missouri Demonstration Project | 10 904 children aged 2-3 years enrolled in the Special Nutrition Program for Women, Infants, and Children in Missouri; consecutive sampling | 1 | SSB intake (servings/day: 0 to <1, 1 to <2, 2 to <3, ≥3); FFQ | Overweight status at follow-up (BMI ≥95th percentile) (OR); measured | Adjusted for age, birth weight, dietary intake, ethnicity, sex, total energy intake; stratified by being normal or under weight at baseline (BMI <85th percentile), being at risk of overweight at baseline (BMI 85th to <95th percentile), or being overweight at baseline (BMI ≥95th percentile) | Children who were at risk for overweight at baseline and consumed ≥1 SSB/day were 1.8-2.0 times more likely to become overweight than children who consumed <1 SSB/day. Children who were overweight at baseline and consumed ≥1 SSB/day were1.8-2.1 times as likely to remain overweight than children who consumed <1 SSB/day. There was a positive but non-significant relation between SSB consumption and development of overweight in children who were normal or under weight at baseline | + |
| Williams (2008),93 USA, Healthy Start Project | 519 children aged 3-4 years, attending 1 of 9 selected preschools in upstate New York; convenience sampling | 4 | Sucrose intake (g/day); direct observation and 24 h recall | BMI at 1 year follow-up (continuous); measured | Adjusted for dietary intake at baseline and follow-up, baseline BMI, ethnicity, sex, total energy intake | Sucrose intake at baseline was inversely associated with BMI at follow-up (β –0.10, P<0.05) | — |
ANOVA=analysis of variance; BF=body fat; FFQ=food frequency questionnaire; FJ=100% fruit juice; OR=odds ratio; SD=standard deviation; SE=standard error; SES=socioeconomic status; SSSD=sugar sweetened soft drinks; SSB=sugar sweetened beverages (including cordials, energy drinks, fruit drinks, iced tea, soft drinks); TV=television; WC waist circumference.
*Most adjusted results are reported unless otherwise stated.
†Higher sugar intake positively associated with weight gain (+), not associated with weight gain (0), and negatively associated with weight gain (−).
‡Fat mass index is usually defined as body fat mass (kg)/height (m2); in Johnson et al98 it is kg/m5.8.